Module metric_groups 

Predefined groupings of the node’s Prometheus metrics.

Most groups are filter-based: they key on the module path of the metrics (preferred over the metric name, because module paths are stable and names are not) and are rendered into a METRICS_FILTER-style string via MetricGroups::to_filter_string.

Each filter-based group is set to a MetricLevel, a verbosity threshold. Individual metrics declare their own level where they are registered, defaulting to MetricLevel::Debug. A group’s level decides which of its metrics are exposed on the metrics endpoint — a metric is exposed when the group’s level is at least as verbose as the metric’s:

• off exposes nothing;
• warn (the group default) exposes only warn-tagged metrics;
• info exposes warn and info metrics;
• debug exposes everything except trace-tagged metrics;
• trace exposes everything.

Metrics whose module belongs to no group are covered by the default threshold (info unless configured), rendered as the leading catch-all directive.

The levels never affect collection: a filter-based group’s metrics are registered and keep collecting regardless of the configured level.

Note the two defaults differ: an untagged metric is exposed from level debug, while a group defaults to the warn threshold, so the default config exposes only the warn-tagged metrics.

The node applies MetricGroups::default() when the config omits metrics.groups entirely, so an omitted and an empty section behave the same.

The hw hardware metrics are registered as a prometheus collector and so bypass the filtering macros entirely. They cannot be level-filtered individually: the hardware group’s level is read once at registration time — MetricLevel::Off skips the whole group, any other level registers it.

Structs

MetricGroups

Per-group verbosity levels for the node’s predefined Prometheus metric groups.

Enums

MetricLevel

Verbosity level for a metric.